If this continues, what I expect to see soon is encryption that decrypts in 2 different unencrypted versions, dependent on the decryption key used.
I can't imagine its that difficult. Make an encryption program that has the option of just normal encryption with 1 decryption key, but the added option of using a second decryption key, and a second set of files. For example: John Doe has 2 harddisks, each with 40 GB of information. One contains the blueprints of the F22 Raptor, the other the complete works from the Gutenberg Project, in 7 different languages. John Doe uses the encryption program to create a single encrypted file, size 80+ GB on a bigger harddisk. If John enters the password Gutenberg, the program decrypts the file into the Gutenberg library. If John enters the password Raptor, he gets the blueprints for the F22. Now law enforcement, if they find the file, not only have to force him to decrypt it, they have to prove there is a SECOND decryption key. If the program uses standard padding of the encrypted file with 100-200% of the original data, they could not even prove that there is a second decryption key just by looking at the size difference between the encrypted file and the Gutenberg library file.